Nat Geo Free Quad Maps

National Geographic updated their system for creating quad maps.  They are easily searchable, free, and can be printed on standard paper.

NatGeoMaps

Census Map

Interactive map of the 2010 Census by the NYT.

Krosnik paper on panel quality

Stanford prof Jon Krosnik has conducted a stream of research on the data quality differences between probability (RDD) samples and non-probability (online panel) samples.  His findings are striking, and call into question the practice of using online panels so heavily in Marketing Research.

Background on the issue at ABC News, and links to his work.  Also, the seminal 2011 AAPOR report on the topic.

Selling Error

You see the TV ads all the time, "People who switch to Company X save on average $480!"  This sounds great, and if you can find a deal like that you should take it if you think it's a good value.  There are a couple reasons this claim is distorted, though, and I'm not talking about the "on average" part--you probably got that already, and it's not that big of a deal anyway.  No, I'm talking about two subtle stories about that magical savings number.

First, is the phrase "people who switch."  I don't know the exact number, so I'll just guess that about 15% of policies are switched a year.  It doesn't matter what the number is, just that it is low.  These switchers will only switch to a company when the quote is good enough.  It could even be a bad quote, but on average the quotes that are accepted will be better.  Each shopper will also probably reject a handful of bad quotes from several companies.  So, that savings number?  Well, it's not the average difference in quotes between Company X and its competitors, it's the average difference of quotes that were accepted (the 15%).  Have you noticed that the savings numbers for different insurance companies are about the same?  That's because these numbers are approximations of the switching price of insurance consumers, which is about the same across the market.  Insurance policies are priced based on actuarial models which contain many inputs, which the "average savings" number averages out, discarding all that detail and leaving the difference at which most people switch.  Not a compelling sales pitch.

The second story involves error, which enters the picture through actuarial models.  An actuarial model is a model that predicts the likelihood of an event through time.  For example, your chance of death rises over time, teenagers are more likely to wreck a car, and so on.  Of course, the model doesn't predict exactly when you will die or which teenager is going to wreck, and the difference between what the model said and what actually happened is called prediction error, or just error.  Insurance companies use this information as an input into their pricing.

Each company has a team of actuaries building models, each making a different prediction about risk, and each setting a different price.  So, when you pop up four rate quotes, you are getting rates set from four different risk models, each with its own errors, higher and low, about how risky you are.  Granted, companies have pricing strategies for certain situations, and some are more discount or full-service, but in general the highly-paid statisticians that work in the industry come to very similar results for individual cases.  However, their drive for individualization can lead to different predictions for the same individual at times.

So how do all these people get all these deals?  Well, there are a large number of people who get quotes and then don't take them.  These transactions are not included in the switcher average.  That leaves the people who did see an offer that was good enough to switch.  This happens because the company is offering a directed discount (new customer, bundling, senior, etc.), or because one company's model made an error and under-priced the situation relative to all the other models.  Think of it as one of the companies making a bad bet and you taking advantage of them.  I don't know how much actual price competition is going on versus variance in model predictions, but I'd wager that many deals are the result of model error.  So, the "average savings" number reflects not only how much of a discount it takes to get people to switch, but also the average difference between your current rate and the lowest prediction by the set of quotes (models) that you got.

If you got a good deal and switched companies, you probably purchased one of these model errors.  You may have gotten an intentional rebate, but you were probably also on the low end of the risk predictions.  That is perfectly fine, and I'm not implying that you or the company you got the policy from are doing anything wrong, or even unusual.

When you see a commercial like this, you should think, "I wonder if I could find a company to give me an accidentally low quote that I could switch to?"  Check out quotes periodically, and you'll probably find that they are a lot closer than on the TV ads, and then remain a non-switcher.  Or snag the occasional great rate.  We use an insurance broker to do this for us, for free.

When I see a commercial like this, I think that it a poor advertising strategy to keep pushing the same overall average number that applies to everyone in the industry.  It's annoying.

Happy error hunting.

IndiHop

This afternoon beer tasting across two different neighborhoods, The Grove and Cherokee is a fun idea.  The connection between these two up-and-coming neighborhoods, or districts really, is that they are both scrappy, urban, and irreverent.  Those are all good things in a neighborhood, in my book.

In parallel, the city is seeing a bubbling up of independent breweries (ugh, call them microbreweries, I guess).  The best thing going right now is a love of hoppy beers, like IPAs.  Here are three current favorites:
Schafly, St. Louis - Dry-Hopped IPA
Charleville, St. Genevieve - Hoptimistic IPA
4 Hands Brewing, St. Louis - Divided Sky Rye IPA

Don't know if any of those will be there, but it will be fun to walk around and see what else is around, and make some new favorites.

Yummy Cranberry Sauce

I know that the name of this dish is contradictory to many people.  Traditional cranberry sauces range from a canned jelly with rings made by the can, to more chutney-like creations where the cranberries are chopped or ground up (my mom makes it this way, hi mom!).  I actually like all versions, but I tried to make one that would have wide appeal--if you don't like this tasty but mild recipe, then let's all just accept that you don't like cranberry sauce and move on.  It's cool.

I set out to make a sauce that had mellow tartness, and really good cranberry flavor.  In this version, the tartness is controlled by using a fair amount of sugar (sugar plus cider instead of water), and by expanding the volume of the dish with the pear.  The flavor is enhanced by the orange zest/juice and the ginger.  The last notable aspect of the recipe is the reduction at the end--this helps make it more viscous, like the jelly from the can, but without the gross.

Here is the recipe.  It makes a fair amount, but it doesn't work well with small quantities.  Freeze the extra and have it on turkey sandwiches later.

Ingredients:
2 cups of apple cider
2 cups sugar
1/2 tsp powdered ginger (or fresh grated if you have it)
2 oranges
2 12 oz bags of cranberries
2 pears

Directions:
In a large pot over medium-high heat, start the cider, sugar, and ginger.  While that is going, stir regularly, and zest and juice the oranges into the pot.  Run the orange over a medium grater to zest, getting most of the good looking orange peel, but not the white pith.  Juice the orange by squeezing it over a small colander to catch the seeds.  Add the cranberries, return to a light boil, and let go until most of them pop, about 15 minutes.

In the meantime, peel, core, and finely chop the pears.  They will disintegrate in the sauce, so chop them fine to speed that process up.  Once the berries are cooked, add the pear and get the heat back up to a slow simmer.  It is important not to go too fast here.  Grab a beer or a glass of wine and stand around and stir this concoction every five minutes or so for up to an hour, reducing the volume by about a quarter to a third.  Test it by seeing how much it sticks to your wooden spoon and make it as thick as you like.

At the end, the sauce should be bright red and have a nice soft jelly consistency.  The only recognizable fruit is the cranberries (and maybe some orange zest).  You should be be able to taste the apple/orange/pear, but they should be secondary to the cranberry.  The ginger should be light enough as not to be noticed.

Hopefully, this sauce is appealing to people, and more folks enjoy cranberries with their turkey!

If you have any ideas for improving this recipe, please comment and let me know.

Mashed Sweet Potatoes with Chipotle in Adobo

I became curious about this dish after seeing someone mention it in a blog comment.  That post mentioned steaming, and didn't say anything about mashing them, but whatever, we're doing a Thanksgiving take here.  The idea is to start with one of the dishes that I always get off the potluck table, sweet potato casserole, and update it with this chile idea.  The result is a sweet, smokey, and warming dish that goes great with a turkey or ham dinner, and is a welcome leftover.  You have to like chiles to like this dish, and you must promise to be sparing with them.

Ingredients:

4 large sweet potatoes

2 chipotle peppers in adobo (1 pepper per 2 potatoes)

2 tsp brown sugar (1 tsp per 2 potatoes)

2 tbsp butter (1 pat per potato)

Directions:

Start a large pot to boil on the stove with enough water to cover the potatoes.  Wash and peel the sweet potatoes, and cut each into 2-3 pieces so they boil faster.  Add the sweet potatoes to the boiling water, return to boil, and boil gently for 20-30 minutes, or until soft enough to mash nicely.

While the potatoes are cooking, open the can of chipotle in adobo (I get mine in the Mexican aisle at Schnuck's or at one of the little Mexican grocers down on Cherokee--not hard to find at all).  Put your pinky in and taste a bit so you know how hot the can you got is relative to your taste--it will be hotter than you think.  Remove only what peppers you need (2) with whatever sauce comes with them, and place them in a small bowl or mortar.  If you think it's going to be too hot, start with 1/2 chile and 1/2 tsp sauce--you can easily add more later.  Mash the peppers with an ice cream scoop or pestle, and remove any skins that won't mash.  The chiles will be dispersed in the potatoes, so you don't want any big chunks.

When the potatoes are cooked, drain them well, and transfer to a bowl.  Add the butter, brown sugar, and chipotles.  Mash everything together with a hand potato masher until all the ingredients are integrated and you have a nice mash.  You're done.

I enjoy this dish about a half-cup at a time, on the side.  Due to the heat, it's not meant to be the main starch.  Hope you enjoy it more often than just the winter holidays.

LaPlace, LA, Andouille Mecca

http://www.nola.com/dining/reviews.ssf?4472?4472

Onion Straws

Inspiration from Fat Johnny.

Wagner of Lambchop on NPR

A wonderful in-studio mini concert by Kurt Wagner, leader of the Nashville collective Lambchop, and one of the great alt-country artists. I love the way he sings.

Costs of War

Link on the costs of war: http://costsofwar.org/

By the Eisenhower Research Project.

Reich Graphic - Limping Middle Class

The NYT graphics team does a great job illustrating the history of class and public policy.

Stereotype Maps

A collection of maps that highlight stereotypes of different parts of the world. You never know whether looking at shit like this helps by shining sunlight on it, or if you're just propagating the stereotypes. There are a couple clever names that give a chuckle. So, there's that.

Charles W. Cushman Photography Library

Charles Weever Cushman, amateur photographer and Indiana University alumnus, bequeathed approximately 14,500 Kodachrome color slides to his alma mater. The photographs in this collection bridge a thirty-two year span from 1938 to 1969, during which time he extensively documented the United States as well as other countries.

- from site

New Shel Silverstein Book

NPR does a story on the new Shel Silverstein book, released by the family.